Timing and data recovery in feed-forward equalization

Abstract

An equalizer is disclosed, and associated operational method. The equalizer has a configuration that balances performance and complexity by obtaining samples that are strongly correlated with future and past transmitted bits, and are weakly correlated with future and past bit transitions, and is useful for timing recovery circuits. Samples are only obtained or collected at time intervals more than one sample period away from the reference sample. Samples are shifted by a delay value less than the sample period, and are obtained at a sample period of one unit interval. A means to adjust the sampling point delay is also disclosed. In an implementation, samples that are within the sample period away from the reference sample are obtained and used for implementing a timing shift, not for equalization of the timing recovery signal. Embodiments are also disclosed for optimizing performance for data recovery.

What is claimed is: 
     
       1. An equalization method, comprising:
 obtaining, by an equalizer, a reference sample of an input signal, the reference sample having a reference sample value; 
 obtaining, by the equalizer, a set of shifted input signal samples at a sample period of one unit interval, the samples being shifted by a delay value, the delay value being less than the sample period, the obtained samples being strongly correlated with future and past transmitted bits and being weakly correlated with future and past bit transitions; 
 multiplying, by the equalizer, each element of said set of samples by a filter coefficient weight to create a set of compensation values; and 
 generating, by the equalizer, an output as a sum of the compensation values and the reference sample value. 
 
     
     
       2. The method of  claim 1  wherein the set of shifted input samples are shifted with respect to an eye-center-aligned timing reference. 
     
     
       3. The method of  claim 1  wherein the set of shifted input samples are shifted with respect to an eye-edge-aligned timing reference. 
     
     
       4. The method of  claim 1  wherein obtaining the set of shifted input samples that are strongly correlated with future and past transmitted bits comprises only obtaining the samples at time instants that are more than the sample period away from the reference sample. 
     
     
       5. The method of  claim 1  wherein:
 the first reference sample of the input signal is taken at instant t, and 
 the set of shifted input samples are taken at instants t+p+nT, 
 where p represents the delay value, which is a fixed timing shift in the range ]0, T[, 
 where T is the period of a data symbol, and 
 where n is an integer and spans a range of values set by the desired filter length, but excludes values −1 and 0. 
 
     
     
       6. The method of  claim 1  wherein the delay value is a fixed value. 
     
     
       7. The method of  claim 1  wherein the delay value is T/2. 
     
     
       8. The method of  claim 1  wherein the delay value is variable and is set to optimize a performance criterion. 
     
     
       9. The method of  claim 1  wherein the delay value is variable and is set to minimize the sum of squares of FFE coefficients. 
     
     
       10. The method of  claim 1  wherein the delay value is variable and is set to minimize the sum of squares of the difference between the output signal and a ideally equalized input signal. 
     
     
       11. The method of  claim 1  further comprising:
 obtaining, by the equalizer, a second set of samples at the sample period of one unit interval, the samples being shifted by the delay value; 
 multiplying, by the equalizer, the second set of samples with a second set of filter coefficients to create a second set of compensation values; 
 adding, by the equalizer, the sum of the second set of compensation values to the output; and 
 adjusting, by the equalizer, the second set of filter coefficients to produce a desired timing point delay. 
 
     
     
       12. An equalization circuit comprising:
 a first signal sampler configured to obtain a reference sample of an input signal, the reference sample having a reference sample value; 
 a second signal sampler configured to obtain a set of shifted input signal samples at a sample period of one unit interval, the samples being shifted by a delay value, the delay value being less than the sample period, the obtained samples being strongly correlated with future and past transmitted bits and being weakly correlated with future and past bit transitions; and 
 a multiplier circuit configured to multiply each element of said set of samples by a filter coefficient weight to create a set of compensation values, and to generate an output as a sum of the compensation values and the reference sample value. 
 
     
     
       13. The equalization circuit of  claim 12  wherein the second signal sampler is configured to only obtain the samples at time instants that are more than the sample period away from the reference sample. 
     
     
       14. The equalization circuit of  claim 12  wherein:
 the first signal sampler is configured to produce signal samples at instants t 0 +nT, 
 where t0 is an initial time reference, n is an integer and T is a symbol period corresponding to the sample period; and 
 the second signal sampler is configured to produce signal samples at instants t 0 +nT+p, 
 where p is a constant in the range ]0, T[. 
 
     
     
       15. The equalization circuit of  claim 12  further comprising:
 a non-transitory machine-readable readable medium storing a set of filter coefficients including a main filter coefficient; and 
 a summing element having one input for each filter coefficient element and producing the output signal; 
 wherein the multiplier circuit comprises: 
 a) a first multiplying element having:
 i) a first input connected to the main filter coefficient; 
 ii) a second input connected to the sample from the first signal sampler; 
 iii) an output connected to one of the inputs of the summing element; 
 
 b) a set of multiplying elements, each having:
 i) a first input connected to one of the filter coefficients; 
 ii) a second input connected to one of the stored samples from the second signal sampler, the stored samples being selected according to the filter length, but omitting samples t 0 +nT+p and t 0 +(n−1)T+p, when processing the sample from the first sampler taken at instant t 0 +nT; and 
 iii) an output connected to one of the inputs of the summing element. 
 
 
     
     
       16. The equalization circuit of  claim 15  wherein:
 the first and second signal samplers are implemented as analog-to-digital converters; and 
 the summing element, the non-transitory machine readable medium, the first multiplying element and the set of multiplying elements are implemented using digital circuits. 
 
     
     
       17. The equalization circuit of  claim 12  wherein the summing element comprises two supplemental inputs, and further comprising:
 two supplemental coefficients set to obtain a desired timing instant variation; and 
 two further multiplying elements, each having:
 i) a first input connected to the stored samples from the second sampler corresponding to samples at instant t 0 +nT+p and t 0 +(n−1)T+p, when processing the sample from the first sampler taken at instant t 0 +nT 
 ii) a second input connected to one of each of said supplemental coefficients; and 
 iii) an output connected to one of each of said supplemental summing node inputs. 
 
 
     
     
       18. An equalization circuit comprising:
 a first signal sampler configured to obtain a reference sample of an input signal, the reference sample having a reference sample value, and to produce signal samples at instants t 0 +nT, where t 0  is an initial time reference, n is an integer and T is a symbol period corresponding to the sample period; and 
 a second signal sampler configured to obtain a set of shifted input signal samples at a sample period of one unit interval, the samples being shifted by a delay value, the delay value being less than the sample period, the obtained samples being strongly correlated with future and past transmitted bits and being weakly correlated with future and past bit transitions, the second signal sampler being configured to produce signal samples at instants t 0 +nT+q, where q is a constant in the range ]−T/2,T/2[; and 
 a multiplier circuit configured to multiply each element of said set of samples by a filter coefficient weight to create a set of compensation values, and to generate an output as a sum of the compensation values and the reference sample value. 
 
     
     
       19. An equalization circuit comprising:
 a. a first signal sampler producing a signal sample at instants t 0 +nT, where t 0  is an initial time reference synchronized close to the edge of each unit interval (UI), n is an integer and T is a symbol period; 
 b. a second signal sampler producing a signal sample at instants t 0 +nT+T/2; 
 c. a set of filter coefficients 
 d. means to store the samples from the first and second signal samplers 
 e. a first summing element having one input for each filter coefficient element and producing a timing recovery output signal 
 f. a first multiplying element having
 i. a first input connected to the main filter coefficient for timing recovery 
 ii. a second input connected to a main sample from the first signal sampler 
 iii. an output connected to one of the inputs of said first summing element for timing recovery 
 
 g. a first set of multiplying elements, each having:
 i. a first input connected to one of said filter coefficients 
 ii. a second input connected to one of the stored samples from the second signal samplers, stored samples are selected according to the filter length, but ignore samples t 0 +nT+T/2 and t 0 +(n−1)T+T/2, when processing the sample from the first sampler taken at instant t 0 +nT 
 iii. an output connected to one of the inputs of said summing element 
 
 h. a second summing element having one input for each filter coefficient element and producing a data recovery output signal 
 i. a second multiplying element having
 i. a first input connected to the main filter coefficient for data recovery 
 ii. a second input connected to a main sample from the second signal sampler 
 iii. an output connected to one of the inputs of said second summing element for data recovery 
 
 j. a second set of multiplying elements, each having:
 i. a first input connected to one of said filter coefficients 
 ii. a second input connected to one of the stored samples from the first signal samplers, stored samples are selected according to the filter length, but ignore samples t 0 +(n+1) T and t 0 +nT, when processing the sample from the first sampler taken at instant t 0 +nT+T/2 
 iii. an output connected to one of the inputs of said second summing element for data recovery. 
 
 
     
     
       20. The circuit of  claim 19  wherein:
 the first and second signal samplers are implemented as analog-to-digital converters; and 
 the summing element, the non-transitory machine readable medium, the first and second multiplying elements and the first and second sets of multiplying elements are implemented using digital circuits. 
 
     
     
       21. The circuit of  claim 19  wherein the summing element comprises two supplemental inputs, and further comprising:
 two supplemental coefficients set to obtain a desired timing instant variation; and 
 two further multiplying elements, each having:
 i. a first input connected to the stored samples from the second sampler corresponding to samples at instant t 0 +nT+T/2 and t 0 +(n−1)T+T/2, when processing the sample from the first sampler taken at instant t 0 +nT 
 ii. a second input connected to one of each of said supplemental coefficients; and 
 iii. an output connected to one of each of said supplemental first summing node inputs.